Sockets in 2e

[FrankMittelbach]

Internal housekeeping

Status of pull request

• Ready to merge

Checklist of required changes before merge will be approved

• Test file(s) added
• Version and date string updated in changed source files
• [n/a] Relevant \changes entries in source included
• Relevant changes.txt updated
• ltnewsX.tex (and/or latexchanges.tex) updated

Basically done, smaller updates and improvements could be added at a later stage.

[u-fischer]

• The output of \ShowSocket looks imho a bit odd:

Socket foo0:
    number of arguments = 0
    available plugs = noop
    current plug = noop
    definition = \protected\long macro:->

! .
\ShowSocket #1->\socket_log:n {#1}\errmessage {}

Where does the !. and the next line come from?

• Why is the default plug for a socket with one argument identity, but for a socket with two or more arguments no-op? (The behaviour imho also contradicts the documentation of \socket_new:nn.

[FrankMittelbach]

Where does the !. and the next line come from?

* Why is the default plug for a socket with one argument `identity`, but for a socket with two or more arguments `no-op`? (The behaviour imho also contradicts the documentation of `\socket_new:nn`.

I can't see that it contradicts, but perhaps the word argument is misleading and I should better use "inputs". The second argument of \socket_new:nn defines how many "inputs" this socket has (the first its name). So a socket with one argument (input) gets identity and you call it as \socket_use:nn {name}{ input }.
For two or more inputs identify makes little sense and as there has to be a default noop is better (and for those without inputs "noop" make sense too.

As to the !.: it says in the documentation the debugging code this is a rough draft and it uses simply \errmessage and I guess that 's where it comes come. Certainly need a better design and a better implementation but I still have hope somebody else is going to do that between now and then.

[u-fischer]

So a socket with one argument (input) gets identity and you call it as \socket_use:nn {name}{ input }.

But \UseSocket is \socket_use:nw and not \socket_use:nn.

For two or more inputs identify makes little sense

why? Isn't #1#2->#1#2 an identity too?
I at least I would have expected that in the following example either both \UseSocket swallow arguments, or that both leave them alone, so either no output, or aaabbbccc:

\documentclass{article}
\NewSocket{foo1}{1}
\NewSocket{foo2}{2}
\begin{document}
\UseSocket{foo1}{aaa}\UseSocket{foo2}{bbb}{ccc}
\end{document}

[FrankMittelbach]

So a socket with one argument (input) gets identity and you call it as \socket_use:nn {name}{ input }.

But \UseSocket is \socket_use:nw and not \socket_use:nn.

that is because with 2e names you just don't know how many arguments follow (not from the command name syntax).
in fact \socket_use:nnn etc is also just \socket_use:nw unless you have checking enabled (but that is only a "future feature and not implemented)

For two or more inputs identify makes little sense

why? Isn't #1#2->#1#2 an identity too? I at least I would have expected that in the following example either both \UseSocket swallow arguments, or that both leave them alone, so either no output, or aaabbbccc:

Then you have to read documentation :-) yes #1#2can be thought of as identity, but in practical terms you would nearly never have any use for it, more likely with noop ... but in any case I would expect we will end up seeing only plugs with 0 or 1 input.

[FrankMittelbach]

one more comment: identity is useful as a default for 1 input (it is its typical use case) but as I claim nearly never for sockets with more inputs. But defining it (unnecessarily) would be a fair amount of work, because it differs for each and every socket whereas noop is trivial to define (and a sensible default in some cases).

[car222222]

Moved to the right place!

No way can things like this be described as “identity” :
#1#2#3#4#5#6#7#8#9

Rather, this is called “concatenation” ;
and it can be very dangerous for some
combinations of “inputs/pins”!! --
Beware of electric shocks, fires, etc!

[u-fischer]

The last lines of the dtx seems to contain forgetten rollback settings and should probably be removed. Apart from this it looks imho fine.

[josephwright]

Looks good to me